It is known that silicon microstructures may be used to form transducers for the sensing of pressure and the like. Typically, a silicon transducer includes a glass substrate having a well formed in one surface thereof. A first surface of a silicon layer is bonded or hermetically sealed to the surface of the glass substrate to form an interior chamber. The opposite exposed second surface of the silicon layer is selectively etched away above the chamber to form a diaphragm of a selected thickness. In a pressure transducer, the diaphragm is deflected in response to the pressure differential across the diaphragm. In a tactile transducer, the diaphragm is deflected in response to a force applied thereto. In either the pressure or tactile transducer, the deflection of the silicon diaphragm within a predetermined deflection range is a linear function of the forces acting on the diaphragm. For example, a 100 micron thick diaphragm may have a deflection range of 0.5-2.0 microns to measure forces between 1-4 pounds.
The transduction may be performed by measuring changes in component values of circuit elements wherein the component values are a function of the deflection of the diaphragm. For example, the interior chamber of the transducer may form a variable gap of a capacitor, the capacitor having a fixed plate at the bottom of the well and a movable plate on the diaphragm. Alternatively, the diaphragm may include piezoresistive elements whose resistive values change as a function of the stress and strain on the diaphragm due to its being deflected.
It is also desirable to put a plurality of such transducers in an array to form a sensing element for robotic fingers. In such an array, it is desirable for each transducer to be addressable. The addressability of each transducer allows the output signal developed by each transducer to be multiplexed onto a data bus for processing by a remote processor. The processor may then determine that an object has been grasped by the fingers, as well as the force the fingers are exerting on the object and the orientation of the object within the fingers.
A disadvantage and limitation of prior art semiconductor tactile transducers utilizing piezoresistive elements is that regions of maximum stress and strain of the diaphragm are a function of the point where force is applied to the diaphragm. To optimize accuracy and linearity of the tactile semicodncutor transducer, it is desirable to place the piezoresistors within the regions of maximum deflection of the diaphragm. Thus, the alignment of the force vector of the force applied to the diaphragm with respect to the position of the piezoresistors is critical. The maximum stress and strain occurring within the diaphragm occurs at the clamp edge of the diaphragm about the periphery of the well. However, the piezoresistors must be offset from the clamp edge. Furthermore, since the piezoresistors are on the diaphragm when the silicon layer is bonded to the substrate, the piezoresistors are difficult to precisely align with respect to the periphery of the well which defines the clamp edge.